Open Research Positions

NEW: Our research group has open research positions for undergraduate students (SOP/LOP/DOP), 9/16 credit UG/MS thesis, summer/winter Interns (at least for 2-3 months), and Higher degree Research Assistants (RP/SAT) (Google form below). These positions are available for BITS students only . Below are some of the possible research projects. All projects are high-impact, allowing participants to perform research and work on real-world problems and data, in collaborations with overseas institutes and leading to research publications or open source software. Positions are often extended over several quarters. We are looking for highly motivated students with any combination of skills: Graph theory fundamentals, machine learning, data mining, network analysis, algorithms, and deep learning.

• Development of Graph Foundational Models
  Building large-scale, general-purpose graph models capable of adapting to diverse downstream tasks across domains.
  
  

• Speech Foundational Models for Indic Languages
  Designing robust speech representation models tailored to the linguistic diversity, phonetic richness, and resource constraints of Indic languages.
  
  

• Visual Reasoning Capabilities of LLMs
  Investigating how large language models interpret and reason over structured visual inputs such as flowcharts, scientific figures, and diagrams.
  
  

• Evaluation of Graph-Inspired Feedback Methods Across Modalities
  Studying how graph-based learning signals and loss functions behave under different data environments, including text, images, and multimodal settings.
  
  

• Development of a scikit-compatible Graph Loss Library
  Creating a plug-and-play graph-inspired loss function library that integrates seamlessly with scikit-learn and mainstream ML pipelines.

• Distinguishing Human-Written vs. AI-Generated Text on Social Media

Exploring methods to differentiate between content authored by humans and text produced by large language models (LLMs) across platforms like LinkedIn, Twitter, and others. 

• ** Designing LLMs playbook using Sora/Gemini (Credits will be provided to the students)

• ** Dataset Curation and Annotation for AI Research, 

• ** Blog writing/survey article on trending CS concepts.

[Last three opportunities ** are also available for early-stage UG students (1st and 2nd year) and can serve as a strong entry point for students looking to grow into deeper research roles.]

📌 We are also open to working on additional related research problems driven by strong student interest and alignment with our broader research themes.

** Recommendation Policy: Should you ask me or not?

I am happy to write letters of recommendation for the students seeking higher study admissions or relevant professional opportunities only if I have had substantial (at least one year or more) continuous academic or research engagement. This ensures that the letter is actually useful for you. The better i know you, the better I will be able speak to aspects of your character and abilities that do not show up on a transcript (e.g. work ethic, intellectual curiosity, collaboration skills, etc.).

If these conditions are not met, I may not be the best person to write a strong recommendation for you, and request you to consider this before reaching out.

** Authorship Policy: I’m happy to include UG/PG researchers as co-authors when their contribution is substantial and sustained throughout the project. In line with IEEE norms, authorship requires meaningful intellectual input, involvement in writing and/or critical revision, and participation through the full paper lifecycle (drafting, experiments, revisions, submission, and publication communication). In particular, experimental contribution is mandatory for empirical papers. If these conditions are not met—such as contributing only briefly, not participating in experiments or writing/revisions, or not being responsive during submission/rebuttal—then I may acknowledge the contribution in the Acknowledgments section instead, which is meant to credit helpful support that does not rise to full authorship.